Because of the high maintenance needed to monitor and filter spammers from the RF Cafe Forums, I decided that it would
be best to just archive the pages to make all the good information posted in the past available for review. It is unfortunate
that the scumbags of the world ruin an otherwise useful venue for people wanting to exchanged useful ideas and views.
It seems that the more formal social media like Facebook pretty much dominate this kind of venue anymore anyway, so if
you would like to post something on RF Cafe's
page, please do.
Below are all of the forum threads, including all
the responses to the original posts.
Post subject: Characteristic impedance silicon
transistor Posted: Tue Feb 27, 2007 11:19 pm
Joined: Tue Feb 27, 2007 6:16 am
I am designing
a UHF transmitter for the sake of learning an interest.
is: Do the surface mount components i.e transistors, I place on the
PCB, take on the characteristic impedance of the PCB or do they have
a characteristic impedance of 50 ohm?
For example, the datasheet
for the BFR182 surface mnt transistor provides S parameters and indicates
that the S parameters provided were measured with a characteristic impedance,
Zo of 50ohm.
The PCB I will be using I calculate to have a 116ohm
It is not practical to design a PCB with a Zo of 50ohm.
Do I calculate the impedances of my transistor using the Zo of 50ohm,
or do I calculate the impedances of the transistor using the PCB Zo
of 116 ohm? Thankyou
S ParametersPosted: Wed Feb 28, 2007 3:27 am
Joined: Wed Feb 22, 2006 3:51 pm
to be missing the relationship between impedance and S parameters.
So, a quick review: to measure impedance, you apply a stimulus
(a voltage or current) and measure the resulting value (a current, if
a voltage is applied, or a voltage if a current is applied).
That's how DC Ohmmeters work, but it gets difficult at VHF and higher
frequencies because we can neither
1. generate an approximately-ideal
voltage or current, nor
2. measure a current or voltage without
The problem is stray capacitance and stray inductance.
For example, a mere 1 pF in parallel at 1 GHz can cause huge errors
in measuring impedances.
So if you can't measure voltages or
currents accurately, what can you measure? The short answer is power.
A slightly longer answer is "reflected or forward power". You do this
using transmission lines rather than plain wires, of course, and it
turns out that we can measure both the power and the angle it's at.
For this measurement to be exact, you need to know the characteristic
impedance of the transmission line. If you use coaxial cable, this is
usually either 50 or 75 Ohms. If you're using microstrip on PCB material,
the characteristic impedance is controllable by the designer, by selecting
the width of the line and the thickness of the material.
for a 50 Ohm system, you need a signal generator with an accurate 50
Ohm output impedance (for the stimulus), and an accurate 50 Ohm load.
And you'll also need one or more directional couplers, and some way
of measuring power.
So suppose that we have a 50 Ohm generator,
connected to 50 Ohm coax, connected to a 75 Ohm resistor, connected
to a 50 Ohm load. The impedance at the resistor will be 75 Ohms in parallel
with 50 Ohms, so where the resistor is connected the impedance will
be less than 50 Ohms.
Where the impedance changes, power is reflected.
You can measure this power with a directional coupler. An Agilent 8510
Vector Network Analyzer ($$$) is a couple of glorified directional couplers,
with a source and power meter.
What the input S parameter is
in a 50 Ohm system, is the reflection coefficient (expressed as a magnitude
and a phase) seen by a 50-Ohm line connected to the transistor input.
You can calculate the actual input impedance if you know S11 and
Z0, the system's characteristic impedance. I posted the formulas in
two recent threads here on RF Cafe-Circuits Forum, the 50-to-75 Ohm
thread, and the amplifier circuit design thread - I won't repeat them
here unless asked.
You make a couple of just-simply-wrong statements
in your posting. Here are some related facts:
1. Designers use
50 Ohm microstrip (=PCB traces) every day of the year - it's not hard
at all. For a calculator, you can go to www.ultracad.com or check any
number of books.
2. PCB's do not have an impedance - the impedance
is a function of the line width and board thickness over the ground
plane. People try to restrict the range to about 25-100 Ohms, but you
can get higher or lower values.
Transistors NEITHER take on the
characteristic impedance of the circuit they're used with, nor do they
themselves have a characteristic impedance. The input and output impedances,
and the forward and reverse gains, are all functions of the operating
voltages and currents.
I hope this helps.
Post subject: Posted:
Wed Feb 28, 2007 9:53 am
Joined: Tue Feb
27, 2007 6:16 am
So to determine
what my input and output impedances of components are, I simply base
my calculation on the given S parameters, input/output reflection coeffs
and convert this to Z parameters and then multiply by the Zo of 50 given
in the datasheets?
What about the transistors legs? Surely the
S parameters of the transistor would change depending on the characteristic
impedance of the PCB it was placed on and the length of the transistors
Ive been using
to calculate my characteristic impedance. The values used are
20 mil track width.
63mil or 1.6mm board thickness.
Zo = 113ohms
If I want a 50ohm characteristic impedance
on this board(pref.)
my track width would have to be 120mil or otherwise
use a 30mil thick board and 50mil wide tracks. Again this track width
is too large.
What values do more experienced designers use?
Different materials for the pcb perhaps?
Your help is very much
Transistor amplifiersPosted: Wed Feb 28, 2007 1:21 pm
Joined: Wed Feb 22, 2006 3:51 pm
Good choice for the microstrip calculator! The folks at UM-Rolla
are pretty sharp, and very helpful.
I'm a bit puzzled, though
- on what grounds is a 50 mil trace "too wide"?
You have two
choices for impedance matching:
1. lumped components (inductors
and capacitors), and
2. transmission lines.
If you're in
the frequency range where you can use lumped components (and with modern
SMT parts, that range is moving upward!), not all traces need to be
50 Ohm microstrip. One "rule of thumb" is that a segment less than 1/10
of a wavelength won't affect things too much. That's one advantage of
computer programs - you can include the interconnections as circuit
elements when you do your analysis. (Still, with modern transistors,
1/10 wavelength may be at a much higher frequency than your chosen operating
frequency, so care is necessary!)
As you surmise, designers often
use other materials - if you read the literature and the on-line material,
you'll find lots of references to materials made by Rogers - ROxxxx,
where xxxx is a number.